01 katz (jk/d) 25/9/01 1:35 pm Page 443

ARTICLE

graphy
Copyright © 2001 SAGE Publications (London, Thousand Oaks, CA and New Delhi)
Vol 2(4): 443–473[1466–1381(200112)2:4;443–473;020413]

From how to why

On luminous description and causal inference in
ethnography (Part I)

■ Jack Katz
University of California, Los Angeles, USA

ABSTRACT ■ Ethnographers often start fieldwork by focusing on

descriptive tasks that will enable them to answer questions about how
social life proceeds, and then they work toward explaining more formally
why patterns appear in their data. Making the transition from ‘how?’ to
‘why?’ can be a dilemma, but the ethnographer’s folk culture provides
facilitating resources for detecting and appreciating especially compelling
descriptions. Recognitions of luminous data often implicitly light the path
to causal inference. In this, the first of a two-part article, three of seven
forms for characterizing the rhetorical effectiveness of ethnographic data
are illustrated and the distinctive resources they offer for causal
explanation are analyzed.

KEY WORDS ■ to come

Over the last 25 years, ethnographic writing has become a major focus in a
debunking critical perspective. Literary devices traditionally used for repre-
senting social life have been rendered deeply suspect by analyses that reveal
how they implicitly construct social realities and subtly claim rhetorical
authority. (Noted early contributions include Gusfield, 1976; Clifford,
1983; Clifford and Marcus, 1986; Van Maanen, 1988.) Although useful for
opening new lines of inquiry, this reflexive turn continues a long-standing
01 katz (jk/d) 25/9/01 1:35 pm Page 444

444 E t h n o g r a p h y 2(4)

failure to appreciate the logic of the culture in which ethnographers rou-

tinely evaluate their data.1 Perspectives for evaluating ethnographic data cut
across writing styles and bridge the anthropology/sociology divide.
However undisciplined and vaguely articulated, a community exists in
everyday discourse about ethnographies. Even while there may be fierce dis-
agreement about the value of given works, there are still common terms for
appreciating ethnographic descriptions as good and for rating given data
passages as especially effective.
To bring into relief the culture for appraising ethnographic depictions,
we should put aside terms of approval such as ‘significant’ and ‘a major con-
tribution’, which slide attention from a focus on the text to an anticipation
of its likely reception. We must also put aside terms such as ‘creative’ and
‘carefully done’, which commend the author’s quality of analysis or pro-
fessional mind. What makes a work ‘interesting’ is often another matter, as
it frequently means that the nature of the theory in a text is counterintuitive
(Davis, 1971). I wish to focus here on terms of appreciation that are
common and distinctively meaningful when applied to the descriptive quali-
ties of ethnographic data.
This culture of appreciation lives in various forms. Readers use it quietly
when reading ethnography, as do authors when writing in anticipation of
readers’ responses. It is easiest to grasp when ethnographies are unself-
consciously characterized with buzz words and rallying cries that praise
descriptions of social life as ‘richly varied’, ‘densely textured’, ‘revealing’,
‘colorful’, ‘vivid’, ‘poignant’, ‘strategic’, or ‘finely nuanced’; for showing
how behavior is ‘crafted’, ‘grounded’, or ‘situated’; or as containing ‘para-
doxes’ and ‘enigmas’ that fascinate the investigator and the reader.
Antonyms are used to dampen enthusiasm and fizzle out calls for support.
Thus poor or weak descriptions are said to be ‘thin’, ‘superficial’,
‘abstracted’, ‘unsurprising’, ‘redundant’, ‘not compelling’, or as presenting
too ‘narrow’, ‘static’, ‘artificial’ or ‘conventional’ a picture of social life.
This culture of appreciation contains neglected practical wisdom for
guiding ethnographic research and writing, especially for meeting the chal-
lenge of moving from ‘how’ to ‘why’, or shifting from a focus on gathering
descriptions of social life to the analytical re-organization of data into
explanatory lines. I do not wish to imply that description is ontologically
separable from explanation. It may well be that when refined sufficiently,
descriptions of how people act merge seamlessly into explanations of their
behavior. The distinction is rooted not in philosophy but in ethnographic
practice. In one common form, the challenge emerges when the researcher
faces a mass of descriptive data and then seeks to address significant debates
that hinge on explanatory problems. Fieldnotes and interviews describing
how people act must be sorted out with relevance for controversies over
why people act as they do. As a practical matter, the path to that degree of
01 katz (jk/d) 25/9/01 1:35 pm Page 445

Katz ■ From how to why 445

refinement where description and explanation merge smoothly is often not

obvious to an ethnographer until a substantial period has been dedicated
self-consciously to description.
It is not easy to avoid this dilemma, for example by gathering data in
the first place literally as answers to ‘why’ questions. Compelling warn-
ings (Mills, 1940; Becker, 1986; Weiss, 1993) alert researchers to the like-
lihood that asking people ‘why’ they have done something will shift the
focus from what the researcher is usually after, which is information on
what subjects did and experienced in times, places, and activities outside
the interview setting. Asking ‘why?’ gives respondents reason to anticipate
that the versions of self they express will be reviewed by the researcher
and whomever else they imagine the researcher communicates with.
Respondents are thus encouraged to exhibit the kinds of comfortable, con-
ventional explanations that presumably will pass muster in the eyes of
these fantasized reviewers. Unless one wants to stick closely to con-
ventional rhetoric or to study artifacts of research situations, the data
resulting from ‘why?’ inquiries are likely to be disappointing. (See also
Geertz, 1979, on natives nonplussed at questions about why they believe
gods are powerful.)
‘How?’ is generally a better way to elicit responses useful for explanation
because it invites a personally historicized, temporally formatted response,
while ‘why?’ authorizes responses formatted in the atemporal and imper-
sonal categories of moral reasoning. Asking someone why they married
someone, chose a residence, or took a job often elicits brief justifications
that highlight present features of the mate, home, or work situation;
features that may well have been discovered since the relationship was
established and that, as current realities, are right at hand to provide an
impressive documentation of the answer. Shifting the question to how one
got that job, found that residence, or got together with that mate com-
monly turns the discussion toward ‘the long story’ that traces how net-
works of social relations and detailed processes of social interaction
worked to shape the respondent’s present status. It is not uncommon for
the researcher to experience a breakthrough toward an explanatory inquiry
specifically by shifting the questions that guide the gathering of data from
‘why?’ to ‘how?’
If we could trust answers to ‘why?’ questions, would we not be in even
worse shape? If research subjects can reliably report why they do the things
we want to understand, who would need us? The question often throws the
burden of explanation onto the subject, reducing our work to, at best,
amplification. The analyst’s answer to ‘why?’ will not necessarily debunk
the member’s, but it will be more of a contribution when it situates and tran-
scends explanations that members conventionally find satisfactory.
The difference between inquiries about how and why is decisive for
01 katz (jk/d) 25/9/01 1:35 pm Page 446

446 E t h n o g r a p h y 2(4)

observational as well as interview practices. When we are taking notes on

what is going on in a given time and place, describing ‘how’ is a reasonable
if never fully completed objective. Data produced through observations in
situ can describe how people work, how they commit crimes, how they
struggle with their intimate domestic relations. But questions about ‘why’
people act as they do logically require observations at different times: 2
events at time 2 can be treated as the result of events at time 1; and they
implicate observations in different places in order to specify the features in
a given situation that were meaningful to and taken into account by the
actor as contingencies of his or her conduct.2 When ethnographers claim to
see why people act as they do without offering comparative data and analy-
sis, they are vulnerable to the charge that they are either invoking magical
powers to ‘see into the minds’ of their subjects, or ‘preaching to the choir’
by applying a pre-set explanatory formula that they presume sentimentally
sympathetic readers will accept.
Explanations always go beyond the descriptions contained in any single
situation represented by a given observational fieldnote, interview segment,
or strip of videotape. An explanation is always an hypothesis. Take, as a
simple example, filling the car with gas. I can describe how I did that on a
given occasion, but why I did it is never really as simple as top-of-the-head
explanations suggest, for example, ‘because I was low on gas’ or ‘because I
needed gas’. I needed gas before I entered the station; I did not rush to a
station the first moment I noticed the gas gauge registering low; and usually
I get there without having to push the car in because it ran completely dry.
In any case my ‘need’ for gas would not explain the extent to which I fill
the tank, nor why I paid with a credit card instead of cash, nor which of the
pumps I chose, nor whether I accept the automatic cut-off as ending the
operation or top up with a final squeeze. As the description of how the act
is conducted improves, the less convincing becomes the initially obvious
answer to ‘why?’
Are we, then, imprisoned in the methodological dilemma that the better
our descriptions, the more problematic their utility for explanation? The
path from describing how people do things to explaining why they so act
must initially be sensed more than articulated, because ethnography eschews
thoroughly pre-fixed designs for data gathering, such as are used for con-
ducting experiments or surveys, in order to discover questions that are
especially salient to its subjects. Yet if the paths from how to why cannot
be marked with bright pointers before ethnographic data gathering is under
way, they often shine at least dimly in the reader’s encounter with the data
themselves. They appear in an initially unreflective appreciation of some
data passages as particularly effective.
In this article and its sequel, seven sets of evaluative terms commonly used
to appraise ethnographic accounts will be reviewed. Underlying each set is
01 katz (jk/d) 25/9/01 1:35 pm Page 447

Katz ■ From how to why 447

a distinct lead toward explanation, the scaffolding of a bridge from ‘how’

to ‘why’.
1. Data effectively set up a problem for explanation when they are resonant
with enigma, paradox, or absurdity.
2. Data are appreciated as strategic in at least two ways that can make data
passages especially valuable. The data set may be bifurcated such that each
datum explicitly supporting the proffered explanation also implicitly
negates a major alternative explanation. Or the flexibility available in par-
ticipant-observation fieldwork may be exploited to give a temporal struc-
ture to the data gathering that parallels the temporal structure of significant
turning points in subjects’ lives.
3. Rich and varied data facilitate ruling out rival hypotheses by specifying defi-
nitions of explanatory problems and qualifying answers.
4. Data are appraised as revealing when they show how forces shaping social
life are routinely overlooked, purposively hidden, or ontologically invisible.
5. Situated data point to mediating conditions that determine whether con-
stantly present possibilities will be mobilized.
6. When data describe how behavior is crafted in ways that are distinctively
aesthetic, colorful, or vivid, they highlight the obdurate characteristics of
social environments, and thus point to the forces that shape social patterns.
7. When data describe poignant moments, they capture people humbled by
transcending concerns that structure persistent patterns in their lives.
It will be assumed that these terms of appreciation represent a culture
that is quite generally used in reading ethnographies. Of course, that is an
empirical argument, and it is not unreasonable to expect that an argument
of this sort would proceed from formal evidence of the range, depth, and
situated mobilization of this evaluative culture. Lacking that base, I trust
that these terms, or at least most of them, are in general currency and will
be familiar to readers.
But even if I am wrong on that count, I offer a logic for shaping ethno-
graphic texts and for at least partially assessing them. The core argument is
that readers’ common-sense evaluative criteria point wisely toward method-
ologically sound links between, on the one hand, the data gathering and
data presentation tasks, which aim at showing how social life takes the
shapes that it does, and, on the other, the explanatory challenge of making
a convincing argument about why social life works as it does. To the extent
these criteria are not now used to appreciate the quality of ethnographic
data, they should be.
Each of the seven sets of criteria highlights a distinct way that ethno-
graphic description can be luminous through subtly promoting the search
for causal explanation. All are induced from ‘tricks of the trade’ (Becker,
1998) by which ethnographers sometimes unwittingly start the move from
01 katz (jk/d) 25/9/01 1:35 pm Page 448

448 E t h n o g r a p h y 2(4)

description to explanation. Because the terms and relative organization of

the seven are not deduced from theory or philosophy, several disclaimers are
in order. First, there is an arbitrariness in the selection of examples, in part
because a particular data passage may fit one, several, or all of the seven.
Second, the examples represent a range of styles and historical periods in
anthropological and sociological ethnography, but every reader will have the
good sense to be stunned in a different way by the limits of my command
of the vast literatures that could be sampled. Despite its length, this essay is
only a beginning.
Third, no claim is made that the seven sets of criteria are exhaustive; the
following argument is not intended as a statement of the logic for assessing
ethnographies. As a whole, these categories of appreciation are most useful
when applied to descriptions of particular people doing specific things, often
with particular other people, in designated places. Many worthy ethnogra-
phies never make the attempt. These include: informant-based models of
how a community typically functions or how some part of society usually
works; structuralist and other interpretations that examine texts indepen-
dent of the social processes by which members create and read them; and
analyses of cultural artifacts as what, in other areas of social science, have
been appreciated as ‘unobtrusive measures’.
I leave for another day the question of whether there are defensible logics
for appreciating ethnographies independent of their value in offering causal
explanation. This is one of a series of essays in which I attempt to construct
a methodology for ethnography that does not require ethnographers to
justify their work with a logic that is either hostile to the causal method-
ologies of fixed design, quantitative research (survey interviews, lab or field
experiments), or that is alien to the logic of humanistic methods (especially
as practiced by historians) (Katz, 1982b, 1997). Causal explanation does
not require positivism in the form of assertions that one can predict, on the
basis of knowing realities at time 1, what will occur at time 2. At a
minimum, ‘explanation’ as I use it here requires ‘retrodiction’, claims that,
if we observe a given phenomenon at time 2, we can state what will have
happened earlier, perhaps in a particular sequence of stages that lead to the
outcome (Katz, in press). Ethnographers may be loathe to put their analy-
ses into the familiar phrases of causal explanation on the common mis-
understanding that causal explanation is inconsistent with an appreciation
of both the emergence of meaning in people’s lives and creativity in cultural
development. In fact, many of the most celebrated ethnographies can be
comfortably digested as retrodictive explanations, such as the description in
Geertz (1972) of all the social alignments that must be put into place to
build the stakes that are entailed when holding a cockfight in Bali, and the
descriptions in Duneier (1999) of the routine negotiations with passers-by,
with guardians of restrooms, with a hostile mayoral administration, and
01 katz (jk/d) 25/9/01 1:35 pm Page 449

Katz ■ From how to why 449

with inconstant co-workers who must be managed in order to sell books

and magazines while living on New York sidewalks. In any case, before
thrashing about in arguments over whether we must or must not bring
causal analysis to ethnography, we should first appreciate how subtly and
usefully we are already doing it.
Part I of this essay treats the first three sets of evaluative terms. Part II,
to be published in the journal’s next issue, treats sets four through seven.

Enigma, paradox, and absurdity

Ethnographers find numerous common solutions to the definition of an ani-

mating problem, including borrowing on a predecessor’s achievements by
returning to a site to test an earlier, celebrated treatment, or examining the
reach of a presumptively relevant existing theory in novel circumstances
(Burawoy, 1991). But if we consider how descriptions of social life them-
selves compel explanatory attentions, one of the most effective ways to move
from a description of how to an organizing focus on why is to present an
enigma, a paradox, or an apparent absurdity. If the sense of mystery is not
peculiar to the ethnographer, he or she can assume that its eventual expla-
nation will also be of general interest.
Fieldworkers will sometimes be stymied in the natural course of data
gathering when they come across a phenomenon that seemingly should have
an obvious meaning but doggedly does not. James Ferguson reports such an
experience in his recent book on Zambian Copperbelt workers and families.
I visited the home of a former mineworker and spotted an unusual phrase
carefully inscribed on an interior wall of the house. The message was near a
windowsill and had been carefully spelled out, in English, in inch-high block
letters. It read, ‘Asia in miniature’. ‘What does it mean?’ I immediately
wanted to know. ‘I don’t know,’ the young man replied, with no great inter-
est. ‘Nothing, really’. But how did it get there? His brother wrote it there, he
explained, a long time ago. It didn’t mean anything. . . . At last a story
emerged. . . . [N]ot yet understanding English, the brother had copied these
words from the caption to a photo in a school atlas. (Ferguson, 1999: 209)
Ferguson reports that this enigma long frustrated a sense of having com-
prehended his subjects well. His description of the phenomenon (what it
expressed, how the phrase got on the wall) works as well to provoke our
curiosity. Of course, the phrase could be dismissed as insignificant, but it
turns out that there was a wealth of non-Zambian, wildly juxtaposed cul-
tural styles that the workers sported in a similarly nonchalant way. Why
adopt all these foreign cultural items if indeed they meant ‘nothing, really’?
Facile explanations, such as that the Zambian urban worker had become
01 katz (jk/d) 25/9/01 1:35 pm Page 450

450 E t h n o g r a p h y 2(4)

fascinated with Western culture, did not go far, in part because the Zam-
bians not only did not appear to care much about understanding Western
culture, but because the items were, in fact, as likely to be Asian or
Caribbean as ‘Western’.
In the end, Ferguson resolved the enigma in a way that substantively
made sociological news but that, as a research experience of solving an
enigma, was not. He determined that the mystery was not due specifically
to his ignorance, nor to intellectual deficiencies in the culture under examin-
ation, but to an ambiguity built into the setting.
[A]s urban workers in the 1950s and 1960s gained a greater ability to live
independently of their rural allies, they became increasingly able to shrug off
the rural-based obligations of wide kin networks, remittances, bridewealth,
visits home, and localized ‘home-folk’ sociality, and along with it (in the long
run) the cultural style that signified the acceptance of these obligations.
(Ferguson, 1999: 231)
By adopting a bizarre mixture of ‘foreign’ styles, Zambian Copperbelt
workers celebrated their (historically temporary) freedom from local ties.
We need not evaluate the substance of Ferguson’s explanation; our
primary concern is to grasp the broadly applicable methodological wisdom
in his research experience. A single data passage describing what a scene
looks like and how it came to be can effectively launch a curiosity to
provide the answer why. An innocently confronted, transparently described
enigma effectively drafts the reader into the explanatory game. The answer
suits the form of traditional social science explanation, although it is not
presented in a formal or abstracted fashion. Ferguson argues that a spon-
taneous, collective fascination with ‘foreign’ styles rose and fell, rising as
ties to home villages were loosened and then, in the undertow of a signifi-
cantly declining economy, giving way to a re-involvement in local cultural
themes.
If Ferguson profited from a sudden confrontation with a neatly packaged,
emphatically underlined question mark conveniently carved into the land-
scape, for others the shift into a search for explanation will arise more
gradually over the course of fieldwork. Jean Briggs was struck by a fasci-
nating enigma only after years of study prepared her to confront a paradox
in what she had come to know of the Inuit. In her early field trips to study
Eskimo shamans, Briggs was challenged by a failure to find any. Then she
was challenged by a sense that something was deeply wrong in her field
relations: she discovered that her subjects discreetly viewed her as impul-
sive, irrational, and aggressive. She explained her interpersonal problems by
viewing the Inuit as governed by expectations of ‘superhuman control of
antisocial moods and superhuman benignity’ (Briggs, 1987: 9), and she
explained their culture in her first book as influenced by values of love/pity
01 katz (jk/d) 25/9/01 1:35 pm Page 451

Katz ■ From how to why 451

for mankind and of rationality (Briggs, 1970). By modeling these values,

and without refusing young children their wishes, the Inuit managed to
produce five-year-olds who were very ‘good’ and ‘undemanding’.
Years later she realized that some patterns she had always observed were
in tension with her previous view of Inuit people as benevolent and non-
aggressively disposed. Indeed, they often seemed sado-masochistic with their
children. There were:
‘games’, little interpersonal dramas, that adults stage all the time with chil-
dren . . . ‘Why don’t you kill your baby brother? Like this!’ [demonstrating];
‘Bite me – harder! Ouch! Isn’t that fun?’; ‘Who’s your daddy? You don’t have
a daddy’. (Briggs, 1987: **)
Note that the paradox is not simply the common experience of finding a
contradiction between patterns in one’s data; it is a tension dramatized in
the subjects’ interaction. That grounded paradox then poses an enigma for
the researcher. Why would a people who are especially distinguished in their
‘superhuman control of antisocial moods and superhuman benignity’
engage in play that encourages their children to be aggressive and that
dramatizes pleasure in inflicting, witnessing, and suffering violence?
Here the question ‘why do they do that?’ emerges vividly when, after the
publication of her first, successful book-length treatment, the ethnographer
reconsiders her original materials and appreciates a neglected pattern of
inter-personal, and intra-personal, conflict. For Briggs, the key phenomenon
is not a conflict of cultures (compare the story of Cohen in Geertz, 1973).
Nor is it a culture of conflict; this is play, and, as ‘just child’s play’ it could
easily be overlooked (compare a dramatization of conflict that is put on
regularly, in designated sites, at high stakes, and for public display, Geertz,
1972). Conflict by itself may be universally fascinating, but it does not make
explanation problematic. Like violence in movies and TV shows, researchers
who describe cultures in conflict and cultures of conflict are likely to draw
attention to their narratives, but as with entertainment conflicts, one can
easily leave a reading without having exercised explanatory curiosity.
Descriptions of conflicts within culture insist more powerfully on answers
to questions why.
Briggs suggests that the moments of dramatized conflict in child-rearing
prepare the way for routine stretches of overtly tranquil living, and that,
conversely, the ambience of sensitively polite social equanimity is a kind of
cultural skin that subtly detects and represses wild, destructive forces. The
analysis is reminiscent of the ‘axes of variation’ that Kai Erikson (1976), in
a few brief but provocative pages, offers as characterizing an Appalachian
community, and as explaining how the social world in Buffalo Creek was
turned upside down by a flood occasioned by negligent mining practices. If
one follows up the theoretical leads suggested (so gently as to be almost
01 katz (jk/d) 25/9/01 1:35 pm Page 452

452 E t h n o g r a p h y 2(4)

unnoticeable) by Erikson and Briggs, ethnographic descriptions of self-con-

flictual social practices should reliably contain far-reaching explanatory keys.
It remains to note how absurdity can also play a powerful role in shift-
ing an ethnographer’s descriptive energies into explanatory channels. (For a
classic anthropological example of an explanation of a ceremony attended
to initially as a striking absurdity, see Bateson, 1938.) The most direct use
of absurdity to set up social research questions has been by the ethno-
methodologists. Continuing a perspective rooted in the writings of Edmund
Husserl, elaborated for social science by Alfred Schutz, and diffused through
American sociology by Harold Garfinkel and Aaron Cicourel, the ethno-
methodological researcher understands that all social life is based on a con-
tinually bootstrapped presumption of inter-subjectivity.
Ethnomethodologists never have developed an ethnographic craft, and
indeed when they consider the matter, they are sometimes hostile.3 The
problem with ethnography that they stumble upon is real. As a matter of
field practice, ethnographers cannot afford not to bridge gaps in under-
standing. Garfinkel’s (1967) students, carrying out his instructions to go
home and, without announcing their purpose, act like a guest or ask family
members to explicate all that was formally ambiguous and implicit in their
speech, could experiment with temporarily disrupting their family relations,
but fieldworkers need to build and maintain rapport in order to gather data.
Ethnographers have the common sense to anticipate that it would not be
propitious to indicate constantly to every person one meets that he or she
does not obviously make sense. Indeed, like anyone using a new language
in everyday-life foreign settings, ethnographers inevitably learn a great deal
about a new social terrain by faking it; that is, by frequently pretending to
understand what they really do not. Although personalities must differ
widely on this score, it is common for ethnographers to exercise substantial
care before pointedly raising specific questions. They fear discovering some-
thing that might be at once wonderful substantively but disastrous practi-
cally, that their subjects have taken deep offense that something beyond
question has been put up for critical inspection. Garfinkel’s students could
afford it if their families reacted with anger or tears because what they
expected to do the next day was to go back to class, not to the field.
If ethnographers are reluctant to create absurd scenes, still, finding a nat-
urally occurring scene that seems absurd, the ethnographer should be
delighted. In it there is likely to be a powerful invitation to try to explain
why members can treat the scene as perfectly sensible. Here is an example
from a study of how lawyers provide civil legal assistance to poor clients.

For at least 25 minutes, L. [lawyer] and Cl. [client] fight over whether L. or
Cl. would go to eviction court and show the judge receipts which, L. claims,
make out a solid defense. Cl. argues that she wouldn’t be paid the same
01 katz (jk/d) 25/9/01 1:35 pm Page 453

Katz ■ From how to why 453

respect by the judge as would L. L. privately insults Cl. by thinking her

unstable (as his comments later show), all the while explaining in a thera-
peutically calm voice, ‘I refuse to insult you by handling this case. You’re fully
capable of doing this yourself.’ Cl. retorts angrily: ‘This is Legal Aid; I’m a
poor person, so I’m your boss; and if I want you to insult me, you’ll insult
me.’ (Katz, 1982a: 221–2).
One might characterize this standoff as a double double bind, a Catch
22 squared. The lawyer in effect asserts that the prospective client only
thinks she needs service; by arguing so powerfully for service, she shows that
she can take care of her problem by herself. His services are available only
to people who are so needy that they do not think that they need his ser-
vices. For her part, the prospective client so forcefully and shrewdly claims
incapacity that she appears capable of taking at least initial steps to resolve
her problem.
Such an event suggests that there are indeed deep gaps between the social
worlds of the workers and clients at this place. Because absurd events are
rare, the social organization of the place must somehow contain a genius
that routinely and invisibly operates to avoid the appearance of gaping
openings. When descriptions are backlit by the strange glow of the absurd,
they pulsate with demands for explanation.4
To review, various types of enigma in ethnographic description guide the
researcher toward a search for explanation. These include: (i) descriptions
of matters in social life that are inexplicable both for the researcher and for
members; (ii) paradoxes that appear as frustrating obstacles when the
researcher attempts to weave descriptive patterns into a coherent narrative;
(iii) self-contradictory actions, like the ‘double binds’ thought by some to
lead to schizophrenia, that are collaboratively sustained; (iv) moments of
absurdity, such as when two people sustain a prolonged interaction by insist-
ing that they cannot. When ethnographers describe the operation of these
enigmas, paradoxes, and little overt lies, they provoke curiosity about the
big sociological ‘why?’: what explains the sense of apparent coherence in
the lives of the people studied? What makes it possible for them to take for
granted that they live in a common social world? Why is social life not
apparently coming apart at the seams constantly?

Strategically organized data

If ethnographers do not collect data according to detailed or inflexible pro-

tocols in research designs, at the outset of a project they can often still antici-
pate explanatory debates they will inevitably enter. In response they may
think strategically about how to organize data collection. In any era of
01 katz (jk/d) 25/9/01 1:35 pm Page 454

454 E t h n o g r a p h y 2(4)

research and in every field, some omnibus explanations will be circulating.

However the research problem is ultimately defined, the ethnographer can
reasonably anticipate explanations that many readers are likely to bring to
the text.
Thus, Lisa Frohmann (1991), in setting out to investigate how gender
discrimination might work its way into the prosecution of rape cases, was
strategic in dividing her participant observation time between prosecution
offices in a substantially affluent, white community, and a low-income
minority city. That her findings were essentially similar in the two offices
enhanced their power. Although ethnographers often appear to abhor the
standard rhetoric of scientific research, what Frohmann did was to try to
control some factors that predictably would be argued as alternative expla-
nations for whatever she found.
Indeed, there were differences in the rape cases handled in the two offices
that reflected the different social ecologies. In upscale ‘Bay City’, the charges
often took the form of ‘date rape’. In low income ‘Center Heights’, rape
cases usually had a gang, prostitution, or drug dealing connection.
Frohmann’s focus, however, was on the process of reviewing and con-
structing cases, and in both offices the prosecutors were involved in essen-
tially the same work of constructing believable stories, not in judging the
truth. They were constrained by the same anticipated problems, which
included ‘ulterior motives’, discrepancies in the accounts given by com-
plainants at different times or to different officials, and variations from
typifications of how sexual assaults presumably occur.
Prosecutors, it turns out, are not so much judges as story tellers. What-
ever their personal values, by and large they are geared toward drafting
scripts that will work down the line, when spun in negotiations with defense
counsel, when recited to grand juries (which decide on charges), or when
dramatized before adjudicating authorities. In effect, the institutional
working environment creates a personality for prosecutors that is substan-
tially independent of gender and community context. They develop an
almost tangible habitus in which interactions with complainants are
managed with a prismatic eye that looks backward to what might have hap-
pened to the complainant, forward to what audiences hearing the tale are
likely to opine, and over the heads of the people immediately involved to
the reputation the prosecutor is building in his or her work circles. The
extreme differences in the social ecologies of the two offices function as a
backlight to every data passage. Each time we read of how a ‘gang’ rape
case was handled in Center Heights, we implicitly look beyond the bound-
aries of the case to how ‘date’ rape cases are handled in Bay City, and vice
versa. It is this double-layer in the reading process that gives the data their
luminosity.
In order to recognize the potential in this methodological strategy, we
01 katz (jk/d) 25/9/01 1:35 pm Page 455

Katz ■ From how to why 455

should note how the data passages become luminous specifically because of
the logical (not racial) blackness in the background. By varying the offices
represented by the data passages included in the text, Frohmann quietly
negates the rival hypothesis that race, poverty, or other community-distin-
guishing characteristics dominate case processing. For example, a case from
Bay City and then another from Center Heights are described as manifest-
ing a discrepancy from a ‘typical’ rape because there was a lack of variety
and experimentation in the sexual activity. In one case the victim described
‘just intercourse’, in the other, ‘all three acts are the same . . . he is grinding
his penis into her butt’ (Frohmann, 1991: 217). Frohmann does not explic-
itly stamp each case as showing that demographics or social ecology do not
control the outcome. And she probably should not, because she does not
want to distract the reader from her positive thesis, which is about the narra-
tive tasks by which prosecutors discredit complainants’ allegations of rape.
But the negative points, the negation of a multitude of reasonable rival
hypotheses, come across, and they are made with even more power because
they are made subtly.
Similarly, the rival hypothesis that male bias leads to rejection of victim
complaints is quietly but powerfully negated by varying the genders of the
prosecutors in the cases described. The upshot is the implication that if there
is a problem of injustice in the ways rape cases are handled by prosecutors,
the problem is deeply implicated in the professional work identity that is
taken on by both female and male prosecutors. The problem is either non-
existent or, as she argues in a later piece, more subtle in that gender, class,
and race bias are ‘inadvertently’ reproduced through the interpretation of
the discrepant moral character of victims’ and jurors’ locales (see Frohmann,
1997, which, however, draws only on Center Heights cases).
It is also useful to bifurcate the data set on the matters to be explained.
Liisa Malkki’s study of Hutu refugees is divided on ‘outcomes’, or what
might be taken as the ‘dependent variable’. Massacres by majority Hutus of
minority Tutsi people, and Tutsi attacks on Hutus, led to mass refugee move-
ments. Some Hutus settled in towns. Many resettled with the aid of inter-
national organizations in designated camps. Malkki documents the Hutus’
identity work by presenting ‘panels’, or lengthy, indented quotations from
individual interviews:
[C]amp refugees saw themselves as a nation in exile, and defined exile, in
turn, as a moral trajectory of trials and tribulations that would ultimately
empower them to reclaim (or create anew) the ‘homeland’ in Burundi. . . .
the town refugees . . . tended to seek ways of assimilating and of inhabiting
multiple, shifting identities – identities derived or ‘borrowed’ from the social
context of the township. (1995: 3)
It is tempting to call Malkki’s study a ‘natural experiment’, but she does
01 katz (jk/d) 25/9/01 1:35 pm Page 456

456 E t h n o g r a p h y 2(4)

not and the temptation is wisely resisted. Not only is there no evidence of
a random assignment of refugees to town or camp, there is also no measure
of pre-refugee identity. It may be that more of the camp refugees had been
involved in slaughters of Tutsi masses and were committed to a more mili-
tant ethnic version of their identities before they arrived in the camps. More-
over, in experiments, not only is the independent variable manipulated by
the researcher, its identity is presumably known before the subjects live out
different fates. In this study, precisely what the ‘independent variable’ is, or
what specifically about town and camp life explain the different identity pro-
cesses, remains to be explored.
Malkki does not set up the study as a causal inquiry into the contingen-
cies of ethnic identification, just as Frohmann did not set up her study as a
test of community ethnic composition, SES, and gender bias in criminal
justice processes. But in both cases, the organization of the data into
moieties relevant to causal debates was strategically useful, given readers’
likely responses. Had Frohmann studied only the poor minority setting and
found bias in case processing, her findings might be dismissed by some
readers as masking race and income biases; had she studied only the more
affluent white setting and found no bias, her findings would have faced
objections that the affluent, politically mobilized community context of the
prosecution office masked gender biases operating elsewhere.
In Malkki’s study, the magnitude of the tragedy behind these refugee
experiences irresistibly draws initial attention. The irony of the historical
events adds interest: international aid established camps to ameliorate the
sufferings that were brought about by impassioned ethnic nationalism, and
camps appear to have recreated the problematic phenomenon. But what sus-
tains interest is not just these compelling outline themes but the additional
irony that, however indisputably tragic the backgrounds of all the refugees,
the tragedy lacked a power sufficient to determine the identities taken on
by the town inhabitants. Something else, something apparently very power-
ful, did.
Is the message of Malkki’s case the depressing power of evil to reinvent
itself, or the inspiring possibility that people can always escape what others
would cruelly make of them? The human freedom implicit in the variation
in the outcomes, despite the blinding horror in the near-background,
compels detailed attention to the evidence of how identities evolved in the
two settings, and underwrites Malkki’s effort to demonstrate the con-
structed nature of the impassioned version of ethnic identity promoted by
the camp refugees. The description of any individual’s outlook in the camp
is made more luminous through the ongoing juxtaposition with outlooks in
the town, and vice versa. In effect, each of the two parts of the data set
frames the other. Whenever camp refugees speak with passion about their
national rights, quotes implying the purposive construction of identity
01 katz (jk/d) 25/9/01 1:35 pm Page 457

Katz ■ From how to why 457

magically appear around the passages due to the reader’s background

appreciation of the relative indifference of town refugees.5
Malkki’s study illustrates an essentially cross-sectional approach to
setting up explanatory questions by documenting simultaneously existing,
non-random contrasting patterns. A complementary appreciation for a data
set may develop when one can describe how identities change over time. The
possibility of developing diachronic data is inherent in virtually all fieldwork
projects, at least to the extent that they involve extended time commitments
to describe a given segment of social life. In most studies, the fit between the
temporal structure of data gathering and the temporal structure of the exam-
ined processes of personal and social change is poor. The ethnographer does
not arrive at the scene at the point at which members first get there, or if he
or she does, the matters of interest develop on time schedules that are incon-
venient to the researcher. Ethnographers use informants not only because
informants possess a cultural perspective or insider’s insight that the
researcher lacks, but also simply because they are there: they were there
before the researcher arrived and they may remain there as the researcher
alternates career phases at the university and in the field.
Ethnographers, particularly those working out of sociology departments,
have often structured the temporal course of their data gathering in paral-
lel with the temporal course of the social processes they have investigated.
This has usually occurred for practical reasons, without any express theor-
etical conviction leading the way. The genre of ‘becoming a . . .’ became
common in American qualitative research at mid-twentieth century, in par-
ticular in the work of the ‘second Chicago School’ of ethnographies carried
out by the students of Everett Hughes and Herbert Blumer. Some of these
studies involved stints of participant observation in which graduate students
would become novice employees in field sites in order to finance their way
through school, a temptation not traditionally open to anthropologists.
In addition, ethnographers in bureaucratically organized societies can
easily find research sites where cohorts of new members are regularly
brought in on a precise calendar schedule and kept together in cohorts for
extended processes of personal transformation: probationary employees,
classes of students, employees in seasonal work. As with rites of passage in
anthropology, the careers of institutionally arranged cohorts offer the
strategic attractions of describing change that runs its course in front of the
ethnographic eye.
Social control agencies, government service administrations, and even
private clubs commonly process people in the form of cases of biographies-
to-be-defined, and often the careers of cases are shorter than the researcher’s
involvement. Thus, within a two-year fieldwork project, one often can reli-
ably project an ability to collect a series of patients screened for admission,
treated, and released; inmates arrested, jailed for months, and sent on to
01 katz (jk/d) 25/9/01 1:35 pm Page 458

458 E t h n o g r a p h y 2(4)

prison; would-be boxers entering a gym, becoming committed amateurs and

then professionals who fight for a purse.6 On a shorter temporal scale, in
many formal work organizations that manage the provision of services, a
sizeable set of cases begins and ends every day, as clients, patients, or cus-
tomers are received, their demands or needs defined, and they are sent off.
Much contemporary service work is billed by the hour and done in gigs, on
service calls, through presentations, and in other work units that have two
methodologically attractive features for the ethnographer who would
describe them. They require worker attentions that are shorter than a day
in length from start to finish, and they require the worker, and sometimes
the client or customer, to remain continually and more or less exclusively
engaged until the ‘job’ is done.
The descriptive challenges of phenomena of these sorts are tempting. The
researcher can expect to describe the course of a large number cases that, in
temporal structure, if not in substance, are homogeneous from the subjects’
perspective. Encountering a phenomenon that appears to have natural
(member-organized) temporal structures, the ethnographer’s descriptive
tasks seem attractively manageable. In turn, the charming appeal of pro-
ducing data that are naturally organized in temporal form has frequently
inspired natural history-like causal explanations.
With Darwin not far in the background, the ‘natural history’ perspective
was a central theoretical perspective in the formative years of American soci-
ology at the University of Chicago (Park and Burgess, 1924). Early studies
of the ‘Chicago school’ often focused on ‘life histories’ and on ‘careers’, for
example Cressey’s (1932) study of the careers of women paid in dance halls
to entertain immigrant men. Howard S. Becker’s dissertation, written under
Everett Hughes, was on ‘The Career of the Chicago Public School Teacher’
(Becker, 1951), and his 1953 article, ‘Becoming a Marijuana User’, became
a widely used research model. Ethnographies of particular settings, such as
social life in a home for unwed mothers (Rains, 1971), carried on the tra-
dition; and theorists drew on multiple ethnographies to conceptualize vast
areas of social life within a framework on transformations of the self, from
studies of personal change in organizational settings (Brim and Wheeler,
1966) to ‘becoming deviant’ in the ubiquitous clutches of state authority
(Matza, 1969). Erving Goffman’s ‘The Moral Career of the Mental Patient’
(1961) was an especially bright model for the study of systematically organ-
ized sequences that forced powerful transformations in personal identity.
Anthropological research has a long tradition of studying rituals and cer-
emonies guiding liminal phases of personal metamorphoses and, perhaps
more recently, historical turning points in personal and collective identities
(Sahlins, 1986; Comaroff and Comaroff, 1992). The parallel topics in the
West have been more elusive because personal transitions are managed by
bureaucratic institutions that profess to overcome any mystery about the
01 katz (jk/d) 25/9/01 1:35 pm Page 459

Katz ■ From how to why 459

contingencies of status passage with explicit criteria, thus contributing to a

culture in which members generally hide their magical beliefs from them-
selves. Thus we think ‘competency’ and ‘ability’ refer to real things, not to
mystical notions, even though these conceptions of essence are never quite
caught in the socially situated conduct that makes up our visible social lives
(see James, 1907).
Yet ontological transformations, or shifts in essential nature, are matters
of intense preoccupation in contemporary lives. We might usefully recall a
common method for teaching a child to ride a bike. Someone runs behind
the would-be adept, stabilizing his balance by holding onto the seat, and after
an unpredictable number of runs and at some unannounced point, the guide
lets go, a fact that the novice often discovers only sometime later, by turning
around. At which point, having learned that he has already mastered the new
competence, he can no longer manage his anxiety and falls down.
Such events become luminous memories because there is something
specifically ineffable at work. Something is at work that eludes vision,
whether the vision is made of the prosaic categories of a rationalized culture
or whether it is a personal glance that would frame and capture an active
force in analytic amber. For virtually any competency valued in contem-
porary society, there are manuals specifying how to acquire or improve one’s
ability. But even when such explicit instructions predictably work, the
experience of acquiring competency has a magical quality that is not itself
cognitively captured. Thus, one can learn to play piano by regularly follow-
ing a course of lessons, but when competency is acquired, the experience
will be of getting a feel for the activity, not of knowing something cogni-
tively but of obtaining new ‘ways of the hand’ (Sudnow, 1978).
If courses of personal development cannot be grasped by subjects cogni-
tively, retrospective interviews are not likely to be satisfactory methods for
studying them. But in their participant and observer roles, ethnographers
can give a temporal structure to data collection that may put them as close
to inherently elusive causal contingencies as any research can. Sociological
ethnographers have repeatedly stumbled upon subjects’ experiences of
magical transformations by finding emotionally powerful frustrations in
their subjects’ efforts to move from performing given lines of action to
taking for granted that they have acquired talents, abilities, or competen-
cies. Instead of writing about magic, liminal states, and taboos that keep
experiences beyond words, sociological ethnographers appear to have
adopted the imagery of existential challenge as the most comfortable
vocabulary for conveying experiences with the ineffable. Here are two
leading examples in contrasting traditions and styles.
When Becker et al. (1961) conducted their now classic observational
study of socialization in medical school, they found the students preoccu-
pied with the problem of how to limit what they should try to learn. So
01 katz (jk/d) 25/9/01 1:35 pm Page 460

460 E t h n o g r a p h y 2(4)

taken were they by what we might call the charisma of medical practice,
and by the teachers whose knowledge seemed both indispensable and ency-
clopedic, the ‘boys in white’ could find no justification for limiting their
studies. It turned out that to the extent that students worked out reasonably
comfortable solutions, it was through relying on thoroughly non-rational,
sentimental means. In particular they discovered the utility of a shared
common culture, for example the domestic culture they shared as married
students or fraternity culture borrowed from days in college. What Becker
and Geer wrote of as ‘latent culture’ provided a collective basis for working
out common limitations on effort (Becker and Geer, 1960). The key contri-
bution of latent culture is that it supplies people in the most rationalized
regions of technologically advanced cultures with that most elusive and
precious quality, trust.
By entering a ghetto gym in Chicago as a participant observer and invest-
ing three years in becoming a boxer, Loïc Wacquant (2000) shaped a stra-
tegic parallel between the temporal structure of his data gathering and the
temporal structure of his own process of development. While most of his
data described others’ careers, his own experience continually provided a
sensitizing perspective on issues of profound personal change. Wacquant’s
data quickly establish that there is something to explain. The gym is a place
of rigorously disciplined civility; the novice boxer comes to see himself in a
new moral profile in contrast to his ghetto counterparts outside. With
respect to practice, the boxer takes on a new habitus, one celebrated in a
newly acquired aesthetic:
Pugilistic beauty resides in the practicalities of the fight itself, not in what it
signifies, as the following comment by Jeff makes obvious: ‘Bein’ able to
deliver a punch jus’ the way you picture it in your head, how you gonna do
it, you know: I mean that’s an art. Jus’ the right timin’, the right speed an’
everythin’, it’s, [raving] it’s a helluva feelin’ after you been trainin’ all the time
an’ hit somebody with that perfect punch’. (Wacquant, 1995: 514 and 2000:
47–60, 67–72, 202–4)
What explains the process of change? Many of the crucial influences go
beyond questions of strong will. How, after all, does a boxer know he has
‘heart’, and when does he know that he has lost it? Why do some come to
assume that they are ‘ready’ for competition while others do not? Critical
turning points occur through participation in the collective world of the
gym. When Wacquant somewhat disingenuously asked whether one could
learn how to box by reading books or could get into shape by working out
at home, he was told emphatically that it would be impossible. His own
experience as a novice boxer and his fieldnotes describing others’ careers
show how interaction with others in the gym guides the process of personal
change in numerous ways. Others are observed as models of styles to
01 katz (jk/d) 25/9/01 1:35 pm Page 461

Katz ■ From how to why 461

imitate; various corporeal rhythms become familiar, thus minimizing poten-

tial surprises in the ring. As colleagues, others pace and push one to exceed
limits, limits which, as a kind of existential mystery, are never objectively
knowable. The older men in the gym, in particular the head coach, Dee Dee,
who arranged matches with promoters, give out measured judgments as to
whether it is likely that given boxers ever will be ready for a fight, and then,
if they have the potential, precisely when it is that they have realized it
(Wacquant, 2000: 99–125, 137–43).
Data are strategically well shaped to locate sociologically significant
phenomena when they track how people move through an anxiously moni-
tored transition from one state to another. Ethnographic data that vividly
describe how people make such a transition are likely to find that the work-
ings of a spiritual, magical, or sentimental culture is a key contingency, one
that explains why some make it and others do not, yet without imputing
determinism. One may never be able to predict which students will develop
a comfortable study regime in school and which will persist in or even
succumb to anxiety, how many poverty lawyers will burnout and how many
will remain committed to the job (Katz, 1982a), or which aspiring boxers
will carry through ‘a project of ontological transcendence . . . to fashion
themselves into a new being. . . .’ (Wacquant, 1995: 507 and 2000: 234–8)7
and which will quit trying, but one may well be able to ‘retrodict’ the experi-
ences and stages an adept will have had to go through, and in that regard,
one can explain the personal and social meanings of the new status.
If we consider the form in which ethnographic description best serves
causal explanatory objectives, we can appreciate the especially strategic
character of a particular variation of temporal data. Sometimes ethnogra-
phers can describe not only how a given phenomenon emerges in social life
but also how it erodes or vanishes. If so, the data will afford extra tests for
causal explanation. Some processes of personal change, and some forms of
collective social transformation, are reversible. Boxers lose heart. Service
workers who distinguished themselves by not burning out when many col-
leagues did will often quit at some later point. Some writers lose the con-
viction that they have anything more to say. The transition from adolescent
to adult is not necessarily eternal, at least not in the sense of securing
respected, empowered, and self-confident status. If our theories of the con-
ditions of making a transition are correct, it should be useful to examine
whether the transition is reversed when those conditions are removed.
Often it will be impractical for the ethnographer to follow the phenom-
enon not only into emergence but also through disappearance. Clearly, the
more microscopic the social phenomenon – for example, when a bout of
laughter or crying is described and analyzed – the more likely the researcher
can sustain a descriptive focus through the life cycle of the phenomenon (see,
for example, the videotaped data in Katz, 1999: chapters 2 and 5). But it is
01 katz (jk/d) 25/9/01 1:35 pm Page 462

462 E t h n o g r a p h y 2(4)

worthwhile remaining alert to the possibility that data gathering may be

stretched to encompass the course of changes out of as well as into a dis-
tinctive state of being. Because ethnographers often cannot specify the
nature of the phenomenon they are looking for before entering the field, they
may not plan to build the temporal structure of the data set strategically.
But when they find that descriptions are shaping themselves to the full life
cycle of a discovered phenomenon, they might appreciate and exploit the
distinctive methodological possibilities for testing emergent explanations.

Rich, contextualized descriptions; varied, massive, densely textured

data sets

Whether they study in sociology or anthropology, ethnographers tend to

develop craft concerns for qualities of description collectively. Some find
models in mentors and colleagues when they are students. In the work of
the ‘Second Chicago School’ there is a common sense of what is well-
presented data, and that sense is very different from the quality of descrip-
tions produced by the ‘first’ Chicago school. I think of the work of the students
of Herbert Blumer and Everett Hughes as distinguished by several enviable
characteristics. Writing had a simplicity and directness that was cultivated in
a generation of students through the years that Helen McGill and Everett
Hughes edited the American Journal of Sociology. Data excerpts feature
impression management in face-to-face interaction, details about the practi-
cal challenges or the ‘work’ of producing conduct, and an eye for idiosyncratic
personal strategies for surmounting commonly encountered problems.
Presumably anthropologists can identify data written in the style of the
Rhodes-Livingstone Institute and Manchester schools. The ‘extended case
method’ (or ‘social dramas’ as developed by Victor Turner, see Kuper, 1973:
184) fit especially well with the study of disputes and conflict resolution.
Gluckman’s law training may be related to the structure he gave to
anthropological data, which is not unlike the structure of knowledge rou-
tinely created by lawyers. Whether or not they play roles in courtroom
dramas, for lawyers a case is a thing with a long history of personal facts
on two or more sides, brought by clients who represent clients who repre-
sent clients (a director of a company that is owned by institutional stock-
holders).
The ‘extended case’ is one described over time and through distinguish-
able stages. It shows the interplay of multiple individuals and social groups
who, through a sustained collective focus, shape cultural themes into an
evolving drama. Within the history of anthropology, data structured into
extended cases stood out against the background of structural functional
ethnography. Typically without describing particular cases in detail,
01 katz (jk/d) 25/9/01 1:35 pm Page 463

Katz ■ From how to why 463

Radcliffe-Brown had offered synchronic understandings of how ways of

doing various things (or economic, kinship, and religious ‘institutions’) fit
together in a community. The diachronic focus of the extended case method
requires the anthropologist to act like a sociological fieldworker engaged in
a ‘case study’. A diffuse and diverse knowledge about the social setting and
history is invoked, but kept as a background resource for documenting the
course of a social phenomenon (an accusation of witchcraft, a property
dispute, the response to a failed fishing expedition or one that was unex-
pectedly successful). The themes that the writer brings to the analytic fore-
ground emerge from a search for patterns across collections of cases.
Pierre Bourdieu has produced a collage style in which ethnographic
materials from various sources describe individual cases in boxes that float
alongside the main text but whose details are not otherwise addressed (e.g.
Bourdieu, 1984). The style is reminiscent of the hyper-boxed layouts of
recent news magazine articles. Bourdieu’s long-term production of the
glossy, contemporary-styled journal, Actes de la Recherche en Sciences
Sociales, must not be coincidental, although the journal was not the origin
but an elaboration of a style he had developed in his earlier monographs.
Bourdieu also favors complex sentence structures – said by some to be
‘French’ or, what is not quite the same thing, ‘Proustian’ – that layer descrip-
tion, self-portrayal, philosophical commentary, and multiple caveats about
how to read a given passage, all compressed with a syntax that the reader
must labor breathlessly to keep running. The whole process requires an
effort analogous to the multi-tasking done effortlessly by computers but
that, when done by a human reader, recreates the extraordinary, even fre-
netic energy that has previously gone into producing the fieldwork and the
text, with the result that the reader obtains a continually refreshed appreci-
ation, when he finally arrives at the end of a sentence and can take a momen-
tary rest, for his tiniest of friends, the period. The point seems to be to
demonstrate a passionately intelligent working style by making the reader
either relive it or drop exhausted in admiration by the wayside as the parade
of evidence and conceptualization moves mercilessly on.
Clifford Geertz’s (1973) ‘thick description’ is helpful in this discussion
because of its celebrity and ambiguity. Although his early monographs were
dense with case data, in his later essays what is ‘thick’ for Geertz is not what
everyone would recognize as description. ‘Thick description’ might lead one
to expect a text with numerous cases illustrating fine variations around a
given theme. In this sense, a ‘thick description’ of the cockfight as an insti-
tution would be built up by presenting evidence on a large number of cock-
fights. The analysis could then make nuanced sense of variations among
them, and would naturally turn to specifying distinctions between cockfights
in one culture and seemingly analogous processes in other cultures such
as cockfights in east Los Angeles and in Mexico City, boxing matches in
01 katz (jk/d) 25/9/01 1:35 pm Page 464

464 E t h n o g r a p h y 2(4)

Las Vegas, soccer games that sometimes turn to destructive madness in

Brazil, ball games played by blacks from urban ghettos at midwestern uni-
versities, who brutally attack each other to enhance the pride of suburban
white alumni contributors, etc. All of these social forms theatrically display
totemic or vicarious battling that is passionately supported by men whose
sexual fantasies, regional loyalties, caste emotions, class interests, and gam-
bling stakes are kept under tenuous guard barely outside the arena of the
distracting entertainment. What Geertz recommends more immediately sug-
gests thick interpretation, a much less happy phrase.
Without describing single cases in detail, with nothing more than an
offhand reference to ‘the dozens of cockfights I saw in Bali’ (1972: 424),
Geertz deconstructs the dramatic meaning of a type of event by finding
behind its typical ways the workings of multiple social hierarchies and oppo-
sitions, such as sexual, religious, village, economic, aesthetic and caste. He
adds to this multiplicity of themes the layers of interpretation that thicken
as subjects interpret each other’s interpretations. Like a psychoanalytic
essay, a case serves to demonstrate the workings of the anthropological sub-
conscious: the author portrays a depth of meaning that the subjects feel, but,
lost in their heavy bets on the matters manifestly at stake, could not readily
state.
Presumably as an unintended consequence, these differences in styles of
presenting data in relation to analysis segregate academic audiences. But
ethnographies in any style may be appreciated when they contain ‘rich and
varied’, ‘contextualized’ or ‘context-sensitive’, and ‘densely textured’ data.
There are at least three useful dimensions buried in these notions. First,
‘richness’, ‘texture’, and ‘contextualization’ capture qualities in the por-
trayal of specific instances. They praise descriptions that show in fine detail
how people as they act take account of and respond to their immediate
social and material landscape. Second, ‘varied’, ‘massive’, and ‘dense’ turn
attention to the data set. They are useful for suggesting that the data as an
aggregation describe a great multiplicity of scenes, actions, or cases. A
crucial third idea also resonates in these phrases: ‘densely textured’ and
‘finely detailed’ suggest that the research focus has been narrowed and made
consistent, such that the data as a whole are characterized by close varia-
tions of similar scenes or lines of action, not widely scattered snapshots of
a vast terrain.
Overall, the appreciation is not unlike the statistical researcher’s satisfac-
tion at seeing the appearance of curves that emerge when they connect clus-
ters of data points on a graph. Earlier, in discussing how fieldwork data can
be organized strategically, it was helpful to think of data sets that are either
split between contrasting versions of plausibly explanatory or explained
phenomena, or that follow instances of transformation over their critical
turning points. Here, it will be helpful to think of fieldwork descriptions that
01 katz (jk/d) 25/9/01 1:35 pm Page 465

Katz ■ From how to why 465

cumulate as dots, which, in the close variation of each from the others, allow
for the perception of emergent figures or themes.
It is commonly remarked that little is written about how researchers turn
from describing fieldwork findings to developing analytically explicit expla-
nations. The problem with excess description is too easily attributed to a
lack of theory in ethnography. The problem routinely is the opposite, an
excess rather than a deficit of explanatory ideas. Initial hypotheses float
around almost all ethnographic projects in the form both of cultural opinion
and as implications from prior studies. As one writes fieldnotes and con-
ducts interviews, so much can seem relevant to so many important debates
that resulting data, despite their enormous collective volume, become scat-
tered and thin. Better to put theory and explanatory concerns aside in the
field in order to describe what seem more or less doings of the same thing,
done again and again, in more or less the same place, by different people.
Writing fieldnotes is usefully practiced as a craft, distinct and separate from
substantive explanatory objectives (Emerson et al., 1995). Then, having
compiled a ‘massive, densely textured, richly varied’ data set, the work of
moving from description to explanation will be a matter of exploiting the
care that was put into description in the first place.
Perhaps the closest that ethnographers have come to specifying a general
approach to mining qualitative data sets for explanatory purposes is ‘ana-
lytic induction’, a methodology which has been used implicitly much more
often than it has been made explicit (Katz, in press). Analytic induction is
continuous with the common sense practices of much scholarship in the
humanities. Researchers do not know quite how to define the thing to be
explained at the outset; they search simultaneously for an explanation that
will fit all the evidence and for a definition of the problem that, without
‘cooking’ or hiding the data, makes relevant only the evidence that fits the
explanation. As one works toward the double fitting of explanatory factors
and the thing to be explained, negative cases give marching orders on a day-
to-day basis.
To illustrate how ‘varied and dense’ data facilitate the development of
causal explanation via analytic induction, consider Donald Roy’s study of
quota restriction in the factory. It took shape as he made detailed records
on his and other workers’ output variations. Industrial sociology had devel-
oped two explanations, neither of which fit what Roy found when he went
to work in a machine shop. One used notions of worker resentment to
explain quota restriction. But Roy found that workers at times ridiculed
each other for not meeting quotas. Another theory would explain quota
restriction by an analysis of economic self-interest: the more work put out,
the more likely the management would revise the piecework system to
require more output for the same pay. But the workers in several ways acted
inconsistently with economic self-interest. At times, they overtly disparaged
01 katz (jk/d) 25/9/01 1:35 pm Page 466

466 E t h n o g r a p h y 2(4)

money motivations. Even after arranging their output so that they could idle
for hours, they sometimes passed up lucrative opportunities to do easy work
on overtime pay scales. And although they would often slow down, such
slow periods were disdained as boring and physically fatiguing. The workers
would often race each other to produce as rapidly as possible with the fewest
errors.
These data were densely detailed. They focused on one area of behavior;
they were not scattered around the factory’s social life. They were richly
varied, showing not only quota restriction but work patterns inconsistent
with that characterization. They were thus nicely prepared for analytic
induction, which strengthens explanations by modifying them through
encounters with negative cases. By shifting his explanatory focus to ‘work
satisfaction’, i.e. by changing his definition of the thing to be explained from
quota restriction, profit maximization and other definitions of the problem,
Roy could make sense of all the rich variations in his data. Work satisfac-
tion meant a lively experience of self-definition. That might require slowing
down or speeding up. As a first order of business, it meant not screwing
oneself by triggering a readjustment of the piece rate, a result dreaded for
its handing over control of personal identity to management. Satisfaction
also meant playing a game, enjoying control over interaction in free time,
expressing aggression against management, and being held in esteem by
close peers. The key drive was not money or resentment, but an interest in
exploiting the work setting so that it gave back vivid news about the self
(Roy, 1953: 513).
Roy never stated his social psychological perspective in terms of self-
exploration. Instead, across his writings, he would emphasize how, in the
variety of settings in which he worked and studied, he was struck by
workers’ passions for gaming, and the power of gaming to make even the
most routine industrial jobs pleasurable.
[I]ndustrial workers . . . have an unflagging interest in games, any old game.
Their thoughts slip easily into the activities of favorite teams. Work times are
often emotionally charged with wagering on whatever is going on at the time.
. . . Betting pools appear on the shop floor with regularity. Here, apparently,
is the drama of life for the modern worker. (Roy, 1953: ??)
While careful to avoid rate busting that would earn the enmity of peers, they
would push themselves for their own benefit:
Sometimes in his sweat and frenzy, arms flailing to blend in mechanical
motion with his spinning machinery, the operator may be hailed by a sym-
pathetic fellow worker: ‘All you need is a broom up your ass and you can
sweep the floor, too.’ (Roy, 1980: 332; see also Roy, 1959–1960)
Roy’s illustrative incident here is luminous to the point of sensational.
01 katz (jk/d) 25/9/01 1:35 pm Page 467

Katz ■ From how to why 467

Does he err by appealing to readers’ non-scientific emotions? Ethnographic

data should systematically over-dramatize the colorful character of people,
their settings and their conduct, for reasons rooted in the logic of empiri-
cally grounding and testing causal explanation. To illustrate I would like to
draw on ‘Pissed Off in LA’, a chapter from How Emotions Work (Katz,
1999: chapter 1) that attempts to explain the emergence and decline of anger
among drivers.
The data are dense in ways that make them at once methodologically
useful and colorful. First, they come from multiple sources: participant
observation, interviews with drivers recounting their recent experiences of
‘road rage’, and, especially useful, incidents recalled from the passenger seat.
Second, their density comes from showing fine variations on behavior within
a specific practical environment, that of driving a car. (That it is a common
environment, one that the reader is likely to know well, does not hurt, so
that he or she may readily add her/his own examples to the data set.) Third,
the events are dense with drama because forceful emotions arise and,
usually, decline in short periods of time. Read as a series they become
tiresome, but each narrative has its moments of compelling interest. This
temporal structure facilitates testing causal theory, making each case,
analytically, multiple cases. Conditions that are hypothesized to lead to
anger, like feeling ‘dumb’ from an inability to get recognized by another,
should be present not only when anger rises but in reverse when anger
subsides, as it regularly does when the offending driver acknowledges his
infraction.
Fourth, I detail personal styles of outfitting cars, of communicating with
other drivers, and of using stereotypes for cursing. I note the ethnicity, age,
gender, class status, and other social characteristics which as a whole indi-
cate that the sample is an extremely diverse set of people. I specify what was
going on with the drivers before they were cut off, where they were going,
what mood they were in, the fantasies they were entertaining, what they
were doing with passengers. All of these ‘colorful’, arguably frivolous details
are there for methodological reasons. They are ways of holding constant
and ruling out rival explanations: for example, that only people of a certain
age, gender, or ethnicity get pissed off when driving; that one must have been
in a rush to be offended when cut off; that one must have had a bad day or
have been frustrated in some regard to make such a big deal out of such a
minor matter.
Finally the variations in experience among the people present are noted.
Passengers, for example, are often in the same traffic, going the same place,
in more or less the same mood as the driver before the incident arises, but
while the driver gets mad, they often get scared or find the event hilarious.
Perspectives from the passenger seat highlight the embodied experience of
driving as a key causal condition. Part of the explanation of ‘pissed off’
01 katz (jk/d) 25/9/01 1:35 pm Page 468

468 E t h n o g r a p h y 2(4)

driving will have to treat seriously the surreal claim that drivers, in driving,
actually become merged with their vehicles. Then, when their cars are cut
off, they are immediately and intimately cut off.
The drama in these data do not provide a good representation of the
experience of driving in Los Angeles in a statistical sense. The ‘color’ of the
descriptions emerges as an unintentional but strategic misrepresentation,
because data gathering and presentation is guided by the methodological
goal of creating not a representative sample of driving but a finely detailed,
massive, finely varied set of descriptions of anger emerging and declining in
a given practical context. The researcher can gather such data by seeking
what appear, in a gut reaction, to be novel cases; data gathering is shaped
almost casually by a lay sensibility to dramatic quality. It is not necessary
or even helpful to think constantly about what scientific evidence requires;
the ethnographer concentrates on getting the facts recorded in as precise
detail as possible. Whether advertently or not, the ethnographer serves the
scientific logic of causal explanation by collecting tightly clustered, finely
differentiated cases, each of which can serve to negate and force revision of
explanatory ideas. The ‘richness’ of the data set refers to the resources it
contains to develop causal explanation.

Acknowledgments

For suggestions I thank Howie Becker, Mitch Duneier, Bob Emerson, Bob Garot,
Laura Miller, Mel Pollner; Gary Fine and the ethnography group at North-
western University: James Ferguson, Steve Lansing, Jean Lave, Liisa Malkki, and
the 2000/2001 culture and history group at the Center for Advanced Study in
the Behavioral Sciences (CASBS). Tom Haskell, David Nirenberg, and Loïc
Wacquant offered especially detailed criticism. Robert Devens and Kathleen
Much improved the writing. The support and isolation at the CASBS was a
necessary blessing.

Notes

Editor’s note: the second part of this article will appear in the next issue of
Ethnography.

1 I sidestep the anthropologists’ debates on the problematic status of writing

on culture, which to some has meant that ‘hanging out’, extensive partici-
pant observation, and the production of reports about ‘life among the so-
and-so’ no longer makes much sense. For sociologists, the focus has not
been on what is useful for writing about culture, much less about ‘cultures’
01 katz (jk/d) 25/9/01 1:35 pm Page 469

Katz ■ From how to why 469

as distinct unities of meaning, but on people and the methodological culture

that is useful for describing people in order to explain their social lives.
‘Culture’ may have lost its moorings in any particular place around which
a participant observer might hang, but it is something else to claim that
people now shape their conduct without reference to the interaction situ-
ation they are in, specifically the practical resources and obstacles they find
at hand for defining how they will be perceived by and thus responded to
by other people who may be around or have access to them. If there is
increased sensitivity to global and virtual culture, there is also, in a simul-
taneous and balancing trend in social thought, increased attention to how
human action is embodied (Bourdieu, 2000). There is no necessary conflict
between these perspectives. ‘The attention to “reading” cultural products
and public representations . . . does not displace but complements the char-
acteristically anthropological emphasis on daily routines and lived experi-
ence’ (Gupta and Ferguson, 1997: 5).
2 Experiments purport to attribute causation by in effect describing behavior
over ‘before’ and ‘after’ times, but they also require control groups, i.e. a
cross-sectional comparison. And if done in labs or other artificial situations,
they will require further cross-sectional comparisons to overcome doubts
about generalizability.
3 See the discussion in Pollner and Emerson (2001), especially the relevant
distinction between early, breaching forms of ethnomethodological studies,
and later studies in which the researcher attempts to become a member. For
a useful demonstration of how one can pursue ethnomethodological ques-
tions with ethnographic data, see Goode (1994).
4 Part of the organization’s solution was to create a special staff that worked
on appellate and class action cases; essentially a staff that only represented
clients they never saw. Other devices are routinely used by lawyers who greet,
interact with, and part from a steady stream of clients every day. One is to
end service interactions by negotiating the appearance of non-endings. A
client will be told some version of ‘your case has substantial merit, and we
could sue on your behalf, but before we can, the judge would require that
we first try x and y, so gather these specific papers, make these specific calls,
and if that does not work out to resolve the problem, come back’. The door,
eternally left open, closes on client after client. No one systematically tracks
whether clients return. When they do not, staff lawyers can comfortably take
for granted that everything worked out to the client’s satisfaction.
5 Although Malkki does not pursue the lead, it should be noted that the bifur-
cation of the data set facilitates causal analysis by highlighting negative
cases of counter-group variations within each population. Examples would
be nationalistic town Hutus or cosmopolitan camp Hutus. The camp
environment institutionalized the tragic past, giving refugees relatively little
to do with regard to their identities but to exploit their commonalities;
01 katz (jk/d) 25/9/01 1:35 pm Page 470

470 E t h n o g r a p h y 2(4)

while the town environment invited diverse ‘strategies of invisibility’, pro-

viding attractive temptations to downplay Hutu identity in mating, resi-
dential location, and occupational engagement. But opportunities are
nowhere distributed with perfect equality, nor are constraints anywhere
experienced with precise similarity. By investigating the biographical and
social interaction processes leading to exceptional, counter-group experi-
ences, the researcher could find clues to the contingencies of nationalistic
(or ‘sedentary’) ethnic identification.
6 For a study of the social organization of qualitative differences among
swimmers that takes advantage of their rapid transformations, see Cham-
bliss (1989: 1): ‘careers in swimming are relatively short; one can achieve
tremendous success in a brief period of time. . . . This allows the researcher
to conduct true longitudinal research in a few short years’.
7 Fighters conceive of boxing not as a springboard for aggression and an exer-
cise in violence but as a skilled bodily trade, a competitive performance craft
requiring sophisticated technical know-how and an abiding moral commit-
ment that will enable them not only to improve their material lot but also,
and more urgently, to construct a publicly recognized, heroic self. Boxing is
the vehicle for a project of ontological transcendence whereby those who
embrace it seek literally to fashion themselves into a new being so as to
escape the common determinations that bear upon them and the social
insignificance to which these determinations condemn them. (Wacquant,
1995: 507)

References

Bateson, G. (1938) Naven. Cambridge: Cambridge University Press.

Becker, H.S. (1951) ‘Role and Career Problems of the Chicago Public School
Teacher’. Unpublished PhD dissertation. University of Chicago.
Becker, H.S. (1953) ‘Becoming a Marihuana User’, American Journal of Soci-
ology 59 (November): 235–42.
Becker, H.S. (1986) Writing for Social Scientists. Chicago: University of Chicago
Press.
Becker, H.S. (1998) Tricks of the Trade. Chicago: University of Chicago Press.
Becker, H.S. and Geer, B. (1960) ‘Latent Culture: A Note On The Theory Of
Latent Social Roles’, Administrative Science Quarterly 5(2): 304–13.
Becker, H.S., Geer, B., Hughes, E. and Strauss, A. (1961) Boys in White: Student
Culture in Medical School. Chicago: University of Chicago Press.
Bourdieu, P. (1984) Distinction: A Social Critique of the Judgement of Taste.
(R. Nice, trans.) Cambridge, MA: Harvard University Press.
Bourdieu, P. (2000) Pascalian Meditations. (R. Nice, trans.) Stanford, CA: Stan-
ford University Press.
01 katz (jk/d) 25/9/01 1:35 pm Page 471

Katz ■ From how to why 471

Briggs, J.L. (1970) Never in Anger: Portrait of an Eskimo Family. Cambridge,

MA: Harvard University Press.
Briggs, J.L. (1987) ‘In Search of Emotional Meaning’, Ethos 15(1): 8–15.
Brim, O.J. and Wheeler, S. (1966) Socialization after Childhood. New York:
Wiley.
Burawoy, M. (ed.) (1991) Ethnography Unbound: Power and Resistance in the
Modern Metropolis. Berkeley: University of California Press.
Clifford, J. (1983) ‘On Ethnographic Authority’, Representations 1(2): 118–46.
Clifford, J. and Marcus, G. E. (eds) (1986) Writing Culture: The Poetics and
Politics of Ethnography. Berkeley: University of California Press.
Comaroff, J. and Comaroff, J. (1992) Ethnography and the Historical Imagin-
ation. Boulder, CO: Westview Press.
Cressey, P.G. (1932) The Taxi-Dance Hall: A Sociological Study in Commer-
cialized Recreation and City Life. Chicago: University of Chicago Press.
Davis, M.S. (1971) ‘That’s Interesting!’, Philosophy of Social Science 7: 309–44.
Duneier, M. (1999) Sidewalk. New York: Farrar, Straus & Giroux.
Emerson, R.M., Fretz, R.I., and Shaw, L. (1995) Writing Ethnographic Field-
notes. Chicago: University of Chicago.
Erikson, K.T. (1976) Everything in its Path: Destruction of Community in the
Buffalo Creek Flood. New York: Simon and Schuster.
Ferguson, J. (1999) Expectations of Modernity: Myths and Meanings of Urban
Life on the Zambian Copperbelt. Berkeley and Los Angeles: University of
California Press.
Frohmann, L. (1991) ‘Discrediting Victims’ Allegations of Sexual Assault: Pros-
ecutorial Accounts of Case Rejections’, Social Problems 38: 213–226.
Frohmann, L. (1997) ‘Convictability and Discordant Locales: Reproducing
Race, Class and Gender Ideologies in Prosecutorial Decisionmaking’, Law &
Society Review 31 (3): 531–55.
Garfinkel, H. (1967) ‘Studies of the Routine Grounds of Everyday Activities’,
in Studies in Ethnomethodology, pp. 35–75. Englewood Cliffs: Prentice-Hall.
Geertz, C. (1972) ‘Deep Play: Notes on the Balinese Cockfight’, Daedalus
(Winter): 1–37.
Geertz, C. (1973) ‘Thick Description: Toward an Interpretive Theory of
Culture’, in The Intepretation of Cultures, pp. 3–30. New York: Basic Books.
Geertz, C. (1979) ‘ “From The Native’s Point Of View”: On the Nature of
Anthropological Understanding’, in Interpretive Social Science, pp. 225–51.
Berkeley: University of California Press.
Goffman, E. (1961) ‘The Moral Career of the Mental Patient’, in Asylums:
Essays on the Social Situation of Mental Patients and Other Inmates,
pp. 125–69. Garden City, NY: Anchor.
Goode, D. (1994) A World Without Words: The Social Construction of Chil-
dren Born Deaf and Blind. Philadelphia: Temple University Press.
Gupta, A. and Ferguson, J. (1997) ‘Culture, Power, Place: Ethnography at the
01 katz (jk/d) 25/9/01 1:35 pm Page 472

472 E t h n o g r a p h y 2(4)

End of an Era’, in Culture, Power, Place: Explorations in Cultural Anthro-

pology, pp. 1–29. Durham: Duke University Press.
Gusfield, J.R. (1976) ‘The Literary Rhetoric of Science: Comedy and Pathos in
Drinking Driver Research’, American Sociological Review 41(1): 16–34.
James. W. (1907) Pragmatism: A New Name for Some Old Ways of Thinking.
New York: Longmans, Green and Company.
Katz, J. (1982a) Poor People’s Lawyers in Transition. New Brunswick, NJ:
Rutgers University Press.
Katz, J. (1982b) ‘A Theory of Qualitative Methodology: The Social System of
Analytic Fieldwork’, in Poor People’s Lawyers in Transition, pp. 197–218.
New Brunswick, NJ: Rutgers University Press.
Katz, J. (1997) ‘Ethnography’s Warrants’, Sociological Methods & Research
25(4): 391–423.
Katz, J. (1999) How Emotions Work. Chicago: University of Chicago Press.
Katz, J. (in press) ‘Analytic Induction’, in N. J. Smelser and P. B. Baltes (eds)
International Encyclopedia of the Social and Behavioral Sciences. Oxford:
Elsevier.
Kuper, A. (1973) Anthropologists and Anthropology: The British School,
1922–1972. New York: Pica Press.
Malkki, L. (1995) Purity and Exile: Violence, Memory, and National Cosmol-
ogy among Hutu Refugees in Tanzania. Chicago: University of Chicago
Press.
Matza, D. (1969) Becoming Deviant. Englewood Cliffs, NJ: Prentice-Hall.
Mills, C.W. (1940) ‘Situated Actions and Vocabularies of Motive’, American
Sociological Review 5(6): 439–52.
Park, R.E and Burgess, E.W. (1924) Introduction to the Science of Sociology.
Chicago: University of Chicago Press.
Pollner, M. and Emerson, R. (2001) ‘Ethnomethodology and Ethnography’, in
P. Atkinson (ed.) Handbook of Ethnography, pp. 118–35. Thousand Oaks:
Sage.
Rains, P. (1971) Becoming an Unwed Mother. Chicago: Aldine-Atherton.
Roy, D. (1953) ‘Work Satisfaction and Social Reward in Quota Achievement’,
American Sociological Review 18 (October): 507–14.
Roy, D. (1980) ‘Manufacturing Consent: Book Review’, Berkeley Journal of
Sociology 25: 329–39.
Roy, D.F. (1959–1960) ‘ “Banana Time”: Job Satisfaction and Informal Inter-
action’, Human Organization 18: 158–68.
Sahlins, M. (1986) Islands of History. Chicago: University of Chicago Press.
Sudnow, D. (1978) Ways of the Hand: The Organization of Improvised
Conduct. Cambridge, MA: Harvard University Press.
Van Maanen, J. (1988) Tales of the Field. Chicago: University of Chicago Press.
Wacquant, L. (1995) ‘The Pugilistic Point of View: How Boxers Think and Feel
About Their Trade’, Theory and Society 24(4) (August): 489–535.
01 katz (jk/d) 25/9/01 1:35 pm Page 473

Katz ■ From how to why 473

Wacquant, L. (2000) Corps et âme: Carnets ethnographiques d’un apprenti

boxeur. Marseilles and Montréal: Agone, Comeau et Nadeau.
Weiss, R.S. (1993) Learning from Strangers: The Art and Method of Qualitative
Interview Studies. New York: The Free Press.

■ JACK KATZ is Professor of Sociology at UCLA. The author of

Poor People’s Lawyers in Transition, Seductions of Crime, and How
Emotions Work, he is currently writing up a team ethnography of
five neighborhoods in Hollywood, California. Address: 2201
Hershey Hall, Box 951551, UCLA, LA, CA 90095-1551, USA.
[email: jack_katz@yahoo.com; jackkatz@soc.ucla.edu] ■
01 katz (jk/d) 25/9/01 1:35 pm Page 474